There are quite a few classic cases in A Corp's report, many involving military, underground intelligence, and telecom fraud, all now being handled with AI. No wonder Jensen Huang mentioned the other day that cybersecurity is a key issue in the AI era. The issues with domestic large models basically just made what everyone already knew public. We'll see how domestic models respond next: giving up distillation will widen the gap, buying their own cards won't keep up with supply, and they might have to go through a low point before bouncing back. Anthropic and OpenAI's technical advantages will continue to expand. Still bullish on cybersecurity and A & O's IPOs; bearish on domestic large models.